1.What is mu (expectation constant) interpreted as when the AR(P) process is non-stationary? 2.Why is it included in the model explicitly? 3.I know that it is the constant mean when the process is stationary. But, what if the parameter values make it non stationary, will it become an unrelated constant?
1. Mu would be there but would be a function of time. 2. Because we like to know the mean. 3. Even if it were a constant then it would still add something to the outcome.